Observed arrival · 2026-09-08
Velrim Turns Documents into Typed Objects
A document-extraction API that maps files into Zod, Pydantic, or JSON Schema-shaped data with field confidence scores and source anchors.
Field notes
Velrim accepts a document plus a developer-defined schema and returns typed output rather than an unstructured extraction blob. Its examples attach confidence values and page-level source anchors to fields, allowing a reviewer to inspect a flagged span instead of the entire document. The homepage shows a five-field invoice schema, a review example with four auto-accepted values and one flagged value, and an eval display reporting 97.8% mean field accuracy against a stated threshold.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue